Research On Network QoS-based AMR Algorithm

The Internet protocol-based voice communications (VoIP) technology has combined traditional public switched telephone network that bases on circuit switching with data communication network that bases on packet switching. Making use of Internet to realize a remote packet voice transmission is cost saving, a wealth of services,easily upgradeable and expandable, etc.It has gradually become the mainstream of communication market. In a VoIP system, real-time voice data compression is needed to reduce network bandwidth requirements. Adaptive Multi-Rate (AMR) coding algorithm provides a variety of different compression bit rates and has become an important VoIP encoder.Meanwhile, AMR has widely used in third generation mobile communication system.In this thesis,the basic principle of AMR codec and the work process of AMR codec are studied.At the same time,the linear prediction analysis,and code-book search algorithm are discussed in detail.On this basis,this thesis proposes an adaptive AMR speech codec algorithm that bases on network QoS parameters estimation. In order to provide better voice communication quality,this algorithm takes advantage of three real-time network QoS parameters which are end to end delay, jitter and packet loss rate to assess the state of network load and adaptively adjust coding rate. Simulation results show that, under different network conditions,improved AMR algorithm improves the user's communication quality.On the other hand, due to the great impact caused by network delays that occur in real-time voice communications,this thesis presents an adaptive jitter buffer algorithm. Delay, jitter and packet loss rate are not mutually independent of each other.A reasonable set of anti-jitter buffer size will optimize the packet loss rate,packet delay and delay jitter of packet voice.Set up the anti-jitter buffer of receiving end adaptively and adjust jitter buffer size based on the current state of network, which make compromise between end to end delay and packet loss rate and achieve better communication quality. Simulation results show that when network delay is less than 400ms and packet loss rate is between 1% and 6%, the jitter buffer size varies between 120-170ms.Through subjective listening tests show that users are satisfied with voice quality.